P1: FIB

July 29, 2000 14:39 AR102 CHAP02

Annu. Rev. Biochem. 2000. 69:31–67Copyright c© 2000 by Annual Reviews. All rights reserved

CRYPTOCHROME: The Second PhotoactivePigment in the Eye and Its Rolein Circadian Photoreception

Aziz SancarDepartment of Biochemistry and Biophysics, University of North Carolina Schoolof Medicine, Chapel Hill, North Carolina 27599-7260; e-mail: AzizSancar@med.unc.edu

Key Words blue-light photoreceptor, flavoprotein, retina, suprachiasmaticnucleus, circadian blind mice, seasonal affective disorder

■ Abstract Circadian rhythms are oscillations in the biochemical, physiological,and behavioral functions of organisms that occur with a periodicity of approximately24 h. They are generated by a molecular clock that is synchronized with the solar dayby environmental photic input. The cryptochromes are the mammalian circadian pho-toreceptors. They absorb light and transmit the electromagnetic signal to the molecularclock using a pterin and flavin adenine dinucleotide (FAD) as chromophore/cofactors,and are evolutionarily conserved and structurally related to the DNA repair enzymephotolyase. Humans and mice have two cryptochrome genes,CRY1andCRY2, thatare differentially expressed in the retina relative to the opsin-based visual photorecep-tors.CRY1is highly expressed with circadian periodicity in the mammalian circadianpacemaker, the suprachiasmatic nucleus (SCN). Mutant mice lacking eitherCry1 orCry2have impaired light induction of the clock genemPer1and have abnormally shortor long intrinsic periods, respectively. The double mutant has normal vision but isdefective inmPer1induction by light and lacks molecular and behavioral rhythmicityin constant darkness. Thus, cryptochromes are photoreceptors and central componentsof the molecular clock. Genetic evidence also shows that cryptochromes are circadianphotoreceptors inDrosophilaandArabidopsis, raising the possibility that they maybe universal circadian photoreceptors. Research on cryptochromes may provide newunderstanding of human diseases such as seasonal affective disorder and delayed sleepphase syndrome.

CONTENTS

HISTORICAL PERSPECTIVE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32CIRCADIAN RHYTHMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33PHOTORECEPTORS IN NATURE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Photoactive Pigments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35Photoreceptors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

CIRCADIAN PHOTORECEPTORS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

0066-4154/00/0707-0031/$14.00 31

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

32 SANCAR

Action Spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Genetic Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Novel Opsins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

ANATOMY OF THE MAMMALIAN CIRCADIAN SYSTEM . . . . . . . . . . . . . . . 46STRUCTURE AND FUNCTION OF MAMMALIAN CRYPTOCHROMES. . . . . . 48

Physical and Biochemical Properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Expression of Cryptochromes in the Retinohypothalamic Axis. . . . . . . . . . . . . . . 50Circadian Oscillation of Cryptochrome Expression. . . . . . . . . . . . . . . . . . . . . . . 51Cellular Localization of Cryptochromes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Interactions of Cryptochromes with Other Clock Proteins. . . . . . . . . . . . . . . . . . 53

GENETICS OF MAMMALIAN CRYPTOCHROMES. . . . . . . . . . . . . . . . . . . . . . 54Phenotype of Cry Mutant Mice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Status of the Molecular Clock in Cryptochrome Mutant Mice. . . . . . . . . . . . . . . 56Cryptochrome Genetics in Other Animals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

MOLECULAR MODEL FOR THE MAMMALIAN CIRCADIAN CLOCK . . . . . . 59CRYPTOCHROMES AND HUMAN HEALTH. . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Seasonal Affective Disorder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Delayed Sleep Phase Syndrome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Jet Lag (Syndrome of Rapid Change in Time Zone). . . . . . . . . . . . . . . . . . . . . . 61Rotating Shift Work. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Circadian Clock and Breast Cancer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

CONCLUDING REMARKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

HISTORICAL PERSPECTIVE

Rhodopsin, the photoreceptor for vision, was discovered in 1877 (1). Its mechanismof action was elucidated by the work of many researchers over a period of morethan a century (2–4). In fact, the last landmark discovery in visual photoreceptionwas the cloning of human genes encoding the blue, green, and red opsins in 1986(5). Because of the rich history of opsin research, the notion that all photosensoryresponses mediated by the eye are initiated by opsins became widely accepted.Thus, the recent discovery that in addition to the vitamin A–based opsins the eyecontains a second, vitamin B2–based pigment, cryptochrome, which is unrelated toopsin and which regulates the circadian clock, was unexpected (6, 7), and initiallythe idea was widely rejected (8).

The existence of a second class of photoreceptors in the eye might have beenpredicted from recent research in circadian rhythms. Animals use light for vision aswell as to sense time of day and adjust their daily behavior (circadian rhythm) ac-cordingly. Data that have been accumulating from circadian research over the past30 years have revealed that the two photosensory systems differ from one anotherwith regard to the manner of integrating the light stimuli, the type of retinal cellsused for absorbing light, and the central nervous system location where the infor-mation is processed. Visual information is processed in the cerebral cortex and usedto construct a three-dimensional representation of the outside world; in contrast,

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 33

photic input into the circadian system is processed in the circadian pacemaker in thehypothalamus to tell the time of day. The issues and views on circadian photorecep-tion have been reviewed recently (9–11) and the molecular aspects of the circadianclock including the important developments of the past three years are covered intwo detailed reviews (12, 13). A recent review on cryptochromes (14) providesa historical perspective on the discovery of these pigments as the photoreceptorinvolved in morphogenesis in plants and circadian photoreception in animals.

CIRCADIAN RHYTHMS

Circadian rhythms are oscillations in the biochemical, physiological, and behav-ioral functions of organisms with a periodicity of approximately 24 h (9, 13, 15).The circadian (from Latincirca = about anddies= day) rhythm is perhaps themost widely observed biological rhythm in nature, conceivably because the ma-jority of organisms are exposed to daily cyclic variation of light (day) and dark(night) and it is advantageous to them to synchronize their physical and behavioralactivities with these cycles. Circadian rhythms are observed in organisms rangingfrom cyanobacteria to humans, and their conservation during evolution suggeststhat they confer a selective advantage. Indeed, it has been experimentally shownthat mutant cyanobacteria with an altered rhythm (16) and ground squirrels withno rhythm (17) were overtaken by their wild-type counterparts either in the testtube or in a simulated field condition. However, the circadian rhythm is not uni-versal: theArchaeaand most of the eubacteria display no circadian rhythm, andseveral model organisms, includingEscherichia coli, Saccharomyces cerevisiae,andSchizosaccharomyces pombe,lack circadian rhythms (13).

Circadian rhythms have three basic features. First, the rhythm is an innate prop-erty of the organism and is maintained under constant environmental conditions. Infact, the circadian rhythm was discovered in 1729 by the Frenchman Jean-Jacquesd’Ortous de Mairan, who found that the daily leaf movements of a heliotrope plantpersisted even when the plant was kept in the dark (18). The length of the innate cir-cadian period varies among species but is quite precise for each species and rangesfrom 22 to 25 h (e.g.Drosophila, 23.6 h; mice, 23.7 h; hamsters, 24.0 h; humans,25.1 h). [A recent study of the period length in humans reports it as 24.2 h (19).]Second, the period length is temperature compensated, so that it is maintained at aconstant value throughout the physiological range of external temperature. Third,circadian rhythms are synchronized with the outside world by light. Althoughheat (20, 21) and other environmental cues can synchronize the rhythm with theenvironment under specific conditions, light is the predominant and perhaps theonly physiologically relevant environmental cue (orzeitgeber,from Germanzeit=time andgeber= giver) for synchronizing the circadian rhythm with the solar day.Figure 1 shows the role of light and dark cycles in regulating activity cycles (pho-toentrainment), and the changes that occur in activity when light is removed fromthe cycle.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

34 SANCAR

Figure 1 Circadian rhythms in mouse and human. This idealized figure shows the dailyoscillation of a physiological variable (melatonin secretion) and a behavioral variable (phys-ical activity) as a function of a cycle of 12 h of light and 12 h of darkness (LD12:12). Thebar at the top shows the light and dark phases where light is turned on at 0600 and turned offat 1800. (Top) Circadian rhythm of plasma melatonin concentration. Note that in both noc-turnal (mouse) and diurnal (human) animals, melatonin levels increase during the night andfall during the day. (Middle) Activity record for mouse. Traditionally the activity recordsare double plotted such that thefirst lineshows activity for the first day on theleft sideandfor the second day on theright side; thesecond lineshows activity for the second day onthe left and the third day on theright, and so on. Plotting the data in this manner facili-tates comparison of successive days both horizontally and vertically.Black barsindicatelocomotor activity. At the end of day 3 the light was turned off for the remainder of theexperiment (DD, indicated byarrow). Under DD, mouse locomotor activity “free-runs”with an intrinsic period of 23.7 h, so the activity phase shifts forward (advances) by about0.3 h each day. (Bottom) Wakefulness record for human.Black barsindicate wakefulness.At the end of the third day the subject was switched to a DD condition. Under DD, humancircadian rhythm free-runs with a period of 25.1 h. As a consequence, upon transition fromLD12:12 to DD the wakefulness phase exhibits a 1-h delay on successive days.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 35

The mechanism by which light is sensed has been a source of great interest inthe field of circadian research. Chronobiologists have searched for the circadianphotopigment using systems of varying complexities. This review includes a briefsurvey of naturally occurring photoreceptors and a detailed analysis of past andcurrent research on circadian photoreceptors in mammals.

PHOTORECEPTORS IN NATURE

The eye is required for circadian photoreception in mammals (9), and until recentlythe only known pigments in the eye were the opsin/retinal-based rhodopsin andcolor opsins. Thus the circadian photoreceptor was assumed to be either an opsinutilized for both vision and circadian entrainment or a special opsin used for circa-dian entrainment only. However, other naturally occurring photoactive pigmentscould function as circadian photoreceptors, especially in plants and protozoa. Al-though there are many naturally occurring light-absorbing compounds, the numberof molecules that convert light energy into either chemical energy (ATP), or in-formation via signal transduction is limited (22). The terms pigment, photoactivepigment, and photoreceptor have been used interchangeably in the literature andwe have followed this common practice in the current review where the contextmakes the meaning clear. Strictly speaking, however, these terms have differentmeanings, detailed next.

Photoactive Pigments

A photoactive pigment is an organic molecule that absorbs in the near UV–visiblelight range and upon absorption of a photon initiates a chemical reaction. It hasbeen argued that a photoactive pigment must fulfill three criteria in order to bephysiologically relevant (22, 23). First, the absorption spectrum of the pigmentshould overlap with the wavelengths that are abundantly represented in sunlight.Second, the pigment must have a high extinction coefficient so that it absorbslight with high efficiency. Finally, the excited state of the photopigment (or thephotoreceptor) must have a long lifetime so that it initiates a photochemical reactionbefore it returns to the ground state by radiationless decay. A list of the currentlyknown photopigments that satisfy one or more of these criteria follows; theirstructures are in Figure 2.

Carotenoids The carotenoids are photoantenna pigments in the photosyntheticsystem and the catalytic pigments in animal and bacterial rhodopsins. Retinal isthe chromophore for the opsin-based visual pigments in animals, and for bacteri-orhodopsin inHalobacteria,which use light energy for phototaxis and to createa proton gradient across the cell membrane and convert light energy into ATPby chemiosmotic coupling. Carotenoids are also found as photochemically inertpigments in carrots, oranges, and pink flamingos.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

36 SANCAR

Figure 2 Photoactive pigments (chromophores). The structures of the chromophoresfound in most photosystems in nature are shown. Retinal-containing photoreceptors ab-sorb in the 350- to 550-nm region. Bilins absorb both in the blue (400–500 nm) and red(600–700 nm) regions. Chlorophylls absorb in the near UV (350–450 nm) and red (600–700 nm) regions. The flavin has an absorption peak at 360 nm in two-electron reduced form;two peaks at 370 and 440 nm in two-electron oxidized form; and peaks at 380, 480, 580,and 625 nm in one-electron reduced (blue neutral radical) form. The unique form of pterin(MTHF) found in the photolyase-cryptochrome family absorbs in the 360- to 420-nm range.

Bilins The bilins are linear tetrapyrroles that function as photoantennas in thelight harvesting complex (LHC) of photosynthetic systems and as the chromophoreof the plant photoreceptor, phytochrome.

Chlorophylls Chlorophylls are cyclic tetrapyrroles and are utilized both as pho-toantennas in the LHC and as the primary photoinduced electron donors in thereaction center (RC) of the photosynthetic systems.

Flavins The flavins are redox-active compounds that are cofactors in many light-independent enzymatic reactions. Flavin adenine dinucleotide (FAD) is the pho-toactive cofactor for the photolyase/blue-light photoreceptor family of proteins(24–30), and flavin mononucleotide (FMN) is the chromophore of the phototropinplant blue-light photoreceptor encoded by theNPH1(nonphototropic hypocotyl)

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 37

gene inArabidopsis(31, 32). Deazaflavin, despite its name, is chemically moresimilar to NAD, which is an obligate one-electron donor/acceptor, than to flavin,which can function as both a one- and a two-electron donor/acceptor (33). 5-Deazariboflavin is found in photolyases from a few species includingCyanobac-teria (34); it functions as a photoantenna in these enzymes (35, 36).

Pterins A special form of pterin, 5,10-methenyltetrahydrofolate (MTHF), is thephotoantenna in the majority of the photolyase/cryptochrome blue-light photore-ceptor family of proteins (6, 29, 37).

Other Potential Photoactive PigmentsThis list of photoactive pigments showsthat few molecules from the vast repertoire of naturally occurring compounds withconjugated bonds and light-absorbing properties are used in photobiological reac-tions. However, this is not necessarily a final list. A few other pigments may alsobe photoactive, even within the narrow range of criteria applied in this article. Forexample, parahydroxycinnamic acid may act as a photoactive pigment in associa-tion with green fluorescent protein (GFP). The parahydroxycinnamic acid cofactorof GFP is excited by either intermolecular energy transfer from aequorin or by di-rect absorption, and it fluoresces in the 500-nm range concomitant withcis-transisomerization of the polypeptide backbone of GFP at the junction with the chro-mophore. Because there is no evidence that this photocycle engenders a chemicalreaction (38), parahydroxycinnamic acid is not included in the list of photoactivepigments. However, a recent report indicates that parahydroxycinnamic acid is thephotoactive pigment of the phytochrome of a photosynthetic bacterium (39).

Photoreceptors

A photoreceptor is an apoprotein containing one or more photoactive pigmentsthat convert light energy into chemical energy or information (i.e. an intracellularsignal). Althoughphotoreceptorandphotoactive pigmentare often used synony-mously, photoactive pigment is actually the chromophore of the photoreceptor.The currently known photoreceptors are listed below.

Rhodopsin Opsin/retinal photoreceptors, along with the photosynthetic system,are the most completely characterized photosystems. Opsins are 30- to 40-kDatransmembrane proteins attached to the chromophore via a Schiff base. In animals,opsins are attached to retinal to form rhodopsin and color opsins, which are the pho-toreceptors for vision. The absorption maxima are affected by the apoprotein se-quence and range from 380 to 560 nm. Humans have four types of opsin/retinal pig-ments in the eye: rhodopsin (λmax= 500 nm), blue opsin (λmax= 426 nm), greenopsin (λmax= 530 nm) and red opsin (λmax= 560 nm). InHalobacteriabacterio-rhodopsin is attached to retininyl aldehyde, and the photoreceptor converts lightenergy into an electrochemical gradient and ultimately into chemical energy in theform of ATP. The primary photochemical reaction in this group of photoreceptors

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

38 SANCAR

is cis to transisomerization of retinal, which in the visual system initiates a signaltransduction cascade through a G protein called transducin. The visual photocycle(Figure 3A) is among the best-understood signal transduction systems.

Photosynthetic System The photosynthetic system consists of the light harvest-ing complex (LHC) plus the photosynthetic reaction center (RC), which containsthe redox-active special chlorophyll pair attached to the RC polypeptides (RC com-plex). The LHC contains hundreds of molecules of antenna pigments includingchlorophylls, carotenoids, and bilins. Because of the multiple pigments involved,photosynthesis employs photons of nearly the entire sunlight spectrum to har-vest energy. The reaction mechanism of the photosynthetic system is known inexquisite detail. A photon absorbed by one of the antenna pigments is transmittedto the RC through the other pigments in the LHC via a series of dipole-dipoleinteractions and eventually excites the special pair, causing photoinduced electrontransfer down one arm of the virtually symmetrical RC complex. This pathwayultimately leads to the splitting of water and the generation of ATP (40).

Phytochrome Phytochromes are cytosolic proteins made up of a homodimer of a125-kDa polypeptide covalently linked to a linear tetrapyrrole (41). They regulatemany plant photoresponses including photomorphogenesis. Phytochromes absorbin the red and far red as well as in the blue region. Light absorption causescis totrans isomerization and converts the photoreceptor from the red-light–absorbingphytochrome (Pr) to the far-red-absorbing phytochrome (Pfr). The plant phy-tochrome is a serine/threonine kinase (42), and the cyanobacterial phytochrome isa histidine kinase (43, 44). Light absorption causes autophosphorylation as wellas phosphorylation of phytochrome kinase substrate 1 (PKS1), suppressor of phyA (SPA1), and phyA-phyB-interacting protein (PIF3) inArabidopsis(45–47). Re-cently it was reported that red light stimulated the binding of phytochrome B toPIF3 (47a) and that red light-activated phytochrome A bound specifically to nu-cleoside diphosphate kinase2 (47b). However, the precise mechanism of signaltransduction by phytochromes is not known.

Phototropin TheNPH1 (nonphototropic hypocotyl) gene encodes the apopro-tein for the photoreceptor for phototropism inArabidopsis. It is a 120-kDaprotein kinase containing FMN and associated with the membrane. It regulates

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 3 Two types of photocycles in nature. (A) Visual photocycle. The primary pho-tochemical reaction is thecis-trans isomerization of retinal by light, which initiates thesignal transduction cascade through the G protein transducin (T). (B) Photolyase photo-cycle. MTHF functions as a photoantenna, absorbing light and transferring the excitationenergy to flavin. The primary photochemical reaction is photoinduced electron transferfrom FADH– to the cyclobutane pyrimidine dimer (substrate), which initiates bond rear-rangement in the dimer and results in generation of two canonical pyrimidines (product)concomitant with restoration of FADHo to FADH– by back electron transfer.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 39

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

40 SANCAR

phototropism in response to blue light and contains two light, oxygen, voltage(LOV) motifs. The LOV repeats in other proteins mediate redox-status-dependentresponses to light, oxygen, and voltage (31, 48). Interestingly, a phytochrome fromthe fernAdiantum capillus-venerishas sequence features of both phytochrome andNPH1 (49); thus, it may function as a super-photoreceptor that regulates responsesto red and blue light. The LOV domains of bothArabidopsisandAdiantumpho-totropins bind FMN stoichiometrically (32) and apparently function as blue-lightsensors.

Photolyase Photolyase is a 55- to 65-kDa protein that repairs UV-induced DNAdamage in a reaction dependent on near UV to blue light (350–450 nm) (50).There are two types of photolyases: one (called photolyase) repairs cyclobutanedipyrimidines and the other [(6-4) photolyase] repairs pyrimidine-pyrimidone (6-4) photoproducts (51). The two types are found in various organisms and exhibit20–30% sequence identity (52–54). The cyclobutane pyrimidine dimer photolyaseis found in many bacteria, someArchaea, and some eukaryotes and a grasshoppervirus (50, 54a). The (6-4) photolyase has not as yet been found in bacteria orAr-chaeabut has been found inDrosophila,Xenopus, rattlesnake, fish andArabidopsis(54). Humans do not have either enzyme (6, 55). Both types of photolyases containtwo photoactive pigments (chromophores). One is invariably FAD in the form ofFADH– (24, 27, 28, 50, 54), and the other so-called second chromophore is a pterin,methenyltetrahydrofolate (28, 37) in most species but 5-deazariboflavin in certainrare species that synthesize this compound (25, 56).

Photolyase repairs DNA as follows (Figure 3B). The damage is recognized in alight-independent manner by the enzyme, which forms a Michaelis complex withthe substrate. Upon exposure of this complex to light, the second chromophoreabsorbs a photon and transfers the excitation energy to the flavin, which in turntransfers an electron to the DNA photoproduct; the cyclobutane ring of the pyrimi-dine dimer or the oxytane ring of the (6-4) photoproduct is broken to generate twopyrimidines (28, 50). Back electron transfer restores the FADHo neutral radical tothe catalytically competent FADH– form, and the enzyme dissociates from DNAto enter new cycles of catalysis (50, 57–59).

Photolyase has certain functional similarities to the photosynthetic system. First,it contains a photoantenna whose sole function is to gather energy and thus increasecatalytic efficiency as measured by the extent of reaction per incident photon. Sec-ond, the catalytic cofactor can be excited by nonradiative energy transfer fromthe photoantenna or by direct absorption of a photon. Third, catalysis is initi-ated by photoinduced electron transfer. Finally, both processes are very efficientwith a quantum yield (number of reaction products per absorbed photon) in therange of 0.7 to 1.0. However, the two systems have significant differences aswell. First, the photosynthetic system is membrane-bound whereas photolyasesare soluble proteins. Second, the photosynthetic system contains hundreds of an-tenna molecules per reaction center whereas photolyases have a single secondchromophore (photoantenna) and a single FAD (catalytic center) per monomeric

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 41

polypeptide. Third, photosynthesis involves a photoinduced electron transfer thatresults in a net oxidation-reduction reaction. In photolyase, repair occurs by cyclicelectron transfer such that at the end of the photocycle the redox states of theenzyme and the substrate/product have not changed (Figure 3B). Finally, photo-synthetic systems use the energy of the entire solar spectrum, whereas photolyasesare UV–blue-light photoreceptors with activity maxima at 370–420 nm for thefolate class and 420–440 nm for the deazaflavin class of enzymes.

Blue-Light Photoreceptors/CryptochromesStudies of the effect of light on or-ganisms have revealed that, remarkably, blue-light responses are universal to nearlyall species tested from bacteria to protists, and from plants to animals. Among theresponses amenable to phenomenological and photophysical studies are photore-activation in bacteria, phototropism and photomorphogenesis in plants, phototaxisin protists, and entrainment of circadian rhythms in fungi andDrosophila(22, 60–63a). Carotene-, flavin-, and pterin-based photoreceptors have been proposed tomediate these photoresponses, but until recently supporting biochemical or geneticdata for most of these claims were not available. A major problem in identifying theblue-light photoreceptors has been that the endpoints of blue-light responses arecellular in nature and not readily amenable to biochemical analysis. A phenomenoncalled light-induced absorbance change (LIAC) in cell-free extracts, which resultsfrom the light-induced photobleaching ofb-type cytochrome absorption, has beenused as an in vitro assay for identifying the blue-light photoreceptors. However, thephysiological significance of LIAC has been controversial (62, 63) and this assayhas not led to identification of any blue-light photoreceptor. The term cryptochromewas coined as a shorthand for blue-light photoreceptors with the following expla-nation: “The pigment system(s) responsible for many of the photoprocesses (asascertained by action spectra) has been nicknamed ‘cryptochrome’ because of [its]importance in cryptogamic plants and its cryptic nature. This term, despised bymany, will suffice us here just because it is shorter than other terms used, such as‘blue (UV) light photoreceptor,’ and it will be a useful term until the pigments areidentified” (60).

By this definition, photolyase, which mediates near UV–blue-light-dependentreversal of far UV effects, was the first cryptochrome to be cloned (64), sequenced(65), purified to homogeneity (24), and characterized (37, 57, 65, 66). An impor-tant reason for the early success in characterization of this blue-light photoreceptoris that in contrast to other blue-light responses, photoreactivation has simple andquantitative in vitro assays (67). Characterization of photolyase revealed an impor-tant fact: many photoreceptors may contain more than one chromophore; hence,any attempt to identify a photoactive pigment by action spectrum measurementsmay be of limited value. Indeed, because the action spectrum maxima of all pho-tolyases are in the 370- to 440-nm range (the two wavelengths where flavin hasabsorption maxima) it was speculated early that the chromophore was a flavin (68).In fact the chromophores of photolyases are flavin and pterin (or deazaflavin) andbecause the flavin is in two-electron reduced (bleached) form (λmax = 360 nm),

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

42 SANCAR

it contributes only 10%–20% to the absorption and action spectrum maxima, whichare dominated by the pterin or deazaflavin (50). In recent years, other “cryp-tochromes” have been cloned and characterized. One is the protein encoded bytheHY4gene ofArabidopsis thaliana(69) and another is the protein encoded bytheNPH1gene of the same organism (31, 46). Both are flavoproteins; NPH1 wasdiscussed briefly above and the FMN-containing photoreceptor encoded byNPH1is now called phototropin (32).

Thus, there are currently three blue-light photoreceptors that have been ge-netically and, to varying degrees, biochemically characterized: photolyase, HY4,and phototropin. They all qualify for the term cryptochrome as originally defined.However, HY4, which has a high sequence homology to photolyase, was called thecryptochrome (30). Two human genes with a high degree of sequence homologyto photolyase and HY4 were discovered (6, 52, 70) and shown to encode proteinsthat have the two photolyase chromophores, FAD and pterin, but no photolyaseactivity (6). These proteins were presumed to mediate a blue-light response such asentrainment of the circadian clock in humans and thus were called cryptochromesas well (6). Hence, cryptochrome has now assumed a precise definition: a photore-ceptor with sequence homology to photolyase but that has no photolyase activityand mediates other blue-light responses. It will be used as such throughout thisreview. A brief summary of plant cryptochromes is presented below. The rest ofthis article will discuss the structure and function of mammalian cryptochromes,which is the main focus of this review.

The first cryptochrome gene was isolated from mustard (Sinapis alba) as asequence homolog of photolyase and was thought to be the gene for the mustardphotolyase (71). Shortly afterward theHY4gene, which encodes one of the twocryptochromes inArabidopsis, was isolated and assigned a function in regulatinghypocotyl elongation in response to blue light (69). This gene was later calledCRY1for cryptochrome (30), and when a second gene homologous to photolyasewas found in theArabidopsisgenome (72, 73) it was namedCRY2(73). Mutationsin CRY2confer a late-flowering phenotype (74). ACRYgene has been isolatedfrom the algaChlamydomonas reinhardtii(75) and fiveCRYgenes were found inthe fernAdiantum capillus-veneris(76). However, there are no genetic data on thefunctions ofCRYs in these organisms.

Plant cryptochromes exhibit about 30% sequence identity with cyclobutanepyrimidine dimer photolyases (69) of microbial origin and 50% sequence identitywith (6-4) photolyase (51). The CRY1 ofArabidopsis thalianaand theSinapis albacryptochrome, which is homologous to CRY2 ofA. thaliana,have both FAD andpterin cofactors but lack photolyase activity (29). Some CRYs, including CRY1and CRY2 ofArabidopsis, have C-terminal extensions of 80–240 amino acids withno homology with photolyases and very little homology among themselves. Thisregion is thought to be involved in effector function, but it is not necessary forCRY activity.

The reaction mechanism of plant cryptochromes is not known. InArabidopsis,both CRY1 and CRY2 appear to be nuclear proteins (14, 74, 76a, 76b) and both bind

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 43

to and are phosphorylated by phytochrome A (a cytosolic protein) in a red-light–dependent manner (77). Significantly, CRY2, but not CRY1, is rapidly degradedby light (blue or red) in a phytochrome-dependent manner (76a, 78). These areimportant clues to the mechanism of photoreception and phototransduction inplants, but at present the pathway from photoreception by cryptochromes to generegulation is not known (14).

Following the discovery that the mammalian cryptochromes are circadian pho-toreceptors (7, 79), it was found that inArabidopsismaintained under dim bluelight, the period of gene expression in aCRY1mutant was increased by about 4 h.This result implicated cryptochromes in circadian photoreception forArabidopsis(80). However, the interpretation of this result is complicated by the contributionof phytochromes to most photoresponses to both blue and red light, includingcircadian response. Thus, the roles of cryptochromes and phytochromes in plantcircadian responses remain to be more precisely defined. A detailed discussionof mammalian circadian photoreceptors and the methods used to identify them ispresented below.

CIRCADIAN PHOTORECEPTORS

Photolyase homologs have been discovered in humans that have a high homologyto Arabidopsiscryptochromes (6, 52, 81) and no photolyase activity (6), whichsuggests that the human cryptochromes might also act as blue-light photorecep-tors (6, 7). In contrast with plants, in which essentially every aspect of developmentand behavior is light regulated, photoregulated physiological responses in humansare limited to vision and photoentrainment of the circadian clock. Hence it wasproposed that hCRY1 and hCRY2 may be circadian photoreceptors (6). Tradition-ally, three approaches—action spectra, genetic analysis, and the search for novelopsins—have been used to identify the circadian photopigments. These are dis-cussed below with particular reference to whether or not existing data are consistentwith the cryptochromes being circadian photoreceptors.

Action Spectra

An action spectrum is a plot of the rate of a photochemical or photobiologicalreaction as a function of the wavelength of light eliciting the reaction. In a simplesystem the action spectrum matches the absorption spectrum of the photoactivepigment. Hence, action spectra have been derived in a number of model systems toidentify the circadian photoreceptor. The circadian response action spectrum wasmeasured inDrosophila(82, 83) using eclosion (pupae emergence) as an endpoint.The results indicated a peak in the 420- to 480-nm region and thus were considered“consistent with the possibility that a carotenoid is the photoreceptor, but theydo not rule out pterins, flavins and other compounds” (82). Recently, the actionspectrum for phase shifting inDrosophilalocomotor activity was determined andessentially the same results were obtained (84).

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

44 SANCAR

The first action spectrum for entrainment of the circadian clock in mammalswas conducted on golden hamsters (85). This study revealed three important fea-tures of the animal circadian photosystem. First, the threshold of light for shiftingcircadian phase (locomotor activity) is higher than that required for eliciting a vi-sual response. Second, the time-dose reciprocity relationship holds for only about3 s for vision; that is, a light dose delivered at high intensity within a millisecondhas the same effect as the same dose delivered at lower intensity over a period ofup to 3 s. In contrast, for circadian photoresponse, as measured by phase shiftingwith light pulses, the time-dose reciprocity relationship held for at least 45 min.Finally, in hamsters, in which the retina contains almost exclusively rods and veryfew cones, the action spectrum maximum was at 500 nm, which is consistent witha rhodopsin-like photopigment as the circadian photoreceptor. However, becausethe visual and the circadian photoresponses showed such a dramatic difference intheir time-dose reciprocity relationship, it was suggested that the retinal cells thatmediate the circadian photoresponse may be different from those involved in imageformation, even though both types of cells might employ the same photoreceptormolecule, rhodopsin (85). Action spectra measurements for phase shifting in micealso revealed a peak near 510 nm, which was ascribed to an opsin-retinaldehydetype of photopigment (86, 87).

Thus, even though the action spectrum inDrosophilaraised the possibility of anon-opsin photopigment, action spectra in mammals were consistent with opsin-based photopigments and not with “cryptochromes,” which absorb in the 370- to440-nm range. However, action spectra on entire organisms are not highly reliablefor identifying photoreceptors for a variety of reasons, including shielding of cer-tain wavelengths by other pigments and light scattering (22, 88). Indeed, geneticand biochemical genetic experiments described below raised different possibilitiesfrom the action spectra regarding the nature of the circadian photoreceptor.

Genetic Analysis

A Neurosporastrain defective in flavin biosynthesis has a severely reduced sensi-tivity to photoentrainment (89), suggesting that the photoreceptor in this organismmay be a flavoprotein, although it is not known whether this organism possessesa cryptochrome. In an interesting experiment,Drosophila raised on an asepticdiet withoutβ-carotene (the precursor of retinal) had essentially no visual re-sponsiveness yet maintained normal circadian photosensitivity (90), leading to theconclusion that the circadian photoreceptor inDrosophilais not opsin-based. In-deed,Drosophilathat are visually blind because of mutations in either rhodopsinbiosynthetic genes or in the phototransduction pathway have nearly normal circa-dian photoentrainment (84, 91), again consistent with a non–opsin-based pigmentbeing the circadian photoreceptor in flies.

From the perspective of mammalian circadian photoreception, the studies onrd mice are especially illuminating. These mice, which include the commonlaboratory strain C3H/HeJ, have a mutation in theβ subunit of rod cGMP

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 45

phosphodiesterase (92). As a consequence of therd/rd genotype, the rods startto degenerate shortly after birth and by 9 weeks of age all rods are lost. The rodloss causes secondary degeneration of cones and at 1 year of age more than 99% ofthe cones have disappeared as well (see Figure 6). When these rodless and conelessmice were tested for circadian functions, their photoentrainment to a cycle of 12 hof light and 12 h of darkness (LD12:12) and their phase shift of locomotor activityby light pulses were indistinguishable from those in wild-type mice (93). Whenthe action spectrum ofrd mice was determined for phase shifting, a maximumwas found near 510 nm, which was ascribed to the middle-wavelength-sensitivecone photoreceptors (M-cones). However, it was also found that light at 357 nmwas more effective than at 510 nm and hence it was concluded that in therdmice, photoentrainment was mediated by a few surviving cones with green- andultraviolet-sensitive opsins (86). Because the number of surviving cones and theamount of detectable retinal in these animals was so low as to be insignificant, itwas also suggested that the mouse retina may contain a novel class of photoreceptorcells with green and UV opsins that are dedicated to circadian photoreception (86).However, in a different strain ofrd mouse (CBA/J), the action spectrum maximumfor phase shifting was 500 nm for the wild type and 480 nm for therd mice, whichled to the conclusion that subsets of both rods and cones contain a unique photopig-ment that works for circadian photoreception but has no role in vision (87, 94).

Finally, the possibility of any role for rods and cones in circadian photorecep-tion was virtually eliminated by the use of transgenic mice in which the rods andcones were selectively destroyed by joining the opsin promoters to the diphtheriatoxin gene (95, 96). Tissue-specific expression during development ablates bothcell types and there are no detectable rod or M-opsin photopigments in these ani-mals (95, 96). Yet circadian responses in these animals measured by a phase shiftin locomotor activity or suppression of melatonin production by 509-nm light wereindistinguishable from those of wild-type animals. Hence, novel opsins located ina different part of the retina were proposed as circadian photoreceptors (95, 97).Apparently, certain blind people with no conscious visual perception of light butnormal circadian photoresponse have retinal lesions analogous to the rodless andconeless mice (98).

Novel Opsins

The normal circadian behavior ofrd mice and the well-established presence ofextraocular circadian photoreceptors in the pineal gland and deep brain of birds(99, 100) and in the pineal gland, deep brain, and parietal eye of reptiles (101) led tothe search for new opsins that might be specialized for the circadian photosystem.These searches have identified several new opsins: pinopsin in the chicken pinealgland (102, 103); melanopsin in the dermal melanophores, the hypothalamus, andthe retina ofXenopus(104); vertebrate ancient opsin (VA-opsin) in some blindrodents and in the amacrine cells of salmon retina (105, 106); and peropsin andencephalopsin in the mammalian brain (107). However, no genetic or biochemical

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

46 SANCAR

data link these opsins to circadian photoentrainment. In particular, with the possi-ble exception of melanopsin no opsin has been found in mammalian retinal cellsother than rod and cone photoreceptors, and the genetic data from mice discussedabove show that circadian photoreceptors are not located in these cells.

ANATOMY OF THE MAMMALIAN CIRCADIAN SYSTEM

As stated above, the circadian system has three fundamental properties: an intrin-sically generated oscillator with a periodicity of approximately 24 h, insensitivityof period length to temperature, and the ability to be adjusted and regulated bylight. Thus, the system is conceived as having three components (108): photore-ceptor (afferent), oscillator (clock), and output (efferent). The synchronization ofthe clock with the solar day is so important that a number of organisms have multi-ple photosensory systems for the circadian input pathway (9, 109, 110). Plants, thedinoflagellateGonylaux polyhedra, andCyanobacteriaappear to have both a redand a blue photoreceptor for setting the circadian clock (80, 110–112). Reptilesand birds have three or four photoreceptor organs (109): the eyes, the parietal eye(reptiles), the pineal gland, and the deep-brain photoreceptors. In mammals, incontrast, all existing evidence indicates that the photoreceptors for both vision andthe circadian clock are located in the eye (10). A recent report on circadian photore-ception through the skin in humans (113) awaits confirmation. In blind hamstersexposure of shaved skin to light does not elicit a circadian response (114).

In mammals total retinal degeneration, enucleation, or severing the optic nervecauses both visual and circadian blindness (96, 98). However, the photorecep-tors for the two systems have different histological locations within the retina(7, 93, 96), and the centers for processing the imaging and circadian photic inputhave different anatomical locations within the brain (115), as shown in Figure 4.Light for vision is absorbed by rhodopsin and color opsins in rods and cones,respectively, which are located in the outer retina near the pigmented epithe-lium. In contrast, light for synchronizing the circadian clock is absorbed by cryp-tochromes in ganglion cells and cells in the inner nuclear layer (amacrine cells,Muller cells, interneurons), which are in the front part of the retina (7) (see Figure6). Signal phototransduction for both systems is through the optic nerve. How-ever, whereas the axons for vision continue their path in the optic nerve to thevisual centers in the cortex, the axons of the circadian system part from theoptic nerve at the optic chiasma and go upward to a pair of neuron clusters inthe anterior hypothalamus called the suprachiasmatic nucleus (SCN). The SCN isthe master circadian pacemaker organ (116).

Although recent research has shown that the circadian oscillator is cell-autono-mous in organisms ranging fromDrosophila (117, 118) to mammals (119), theSCN is the master pacemaker that apparently overrides all local oscillators andcoordinates the peripheral oscillators with the pacemaker clock. An independentclock exists in the retina, but in the absence of the SCN it cannot act as a pacemaker

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 47

Figure 4 The retinohypothalamic tract for circadian phototransduction in mammals. Lightabsorbed by the rods and cones generates a signal that is transmitted through the optic nervesto the visual cortex (green). Light absorbed by the neurons in the inner nuclear layer andganglion cell layers of the retina produces a signal that travels through the optic nerve to thesuprachiasmatic nucleus (red), the circadian pacemaker above the optic chiasma. (Adaptedfrom Reference 115.)

to regulate the organism’s circadian rhythm (120). Retrograde labeling experimentshave shown that a small subset of the retinal ganglion cells form the basis of theretinohypothalamic tract that transmits photic information from the eye to theSCN. Remarkably, cell lines derived from the rat SCN maintained a robust circa-dian rhythm as measured by the uptake of the metabolic marker 2-deoxyglucose.Most significant, the transplantation of SCN cell lines, but not mesencephalic orfibroblast lines, restored the circadian activity of SCN-lesioned rats (121). Theseproperties clearly explain the unique pacemaker function of the SCN as opposedto other tissues that also have circadian oscillator systems. Similarly, even thoughtranscription of the rat period 2 gene (rPER2) oscillates with circadian period-icity in both the SCN and peripheral tissues, destroying the SCN abolished theperipheral oscillation of therPER2 transcription, leading to the conclusion thatSCN-released humoral factors control the peripheral oscillators (122). The mech-anism by which the SCN itself is synchronized with the outside world throughcryptochromes is discussed below.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

48 SANCAR

STRUCTURE AND FUNCTION OF MAMMALIANCRYPTOCHROMES

Physical and Biochemical Properties

Cryptochromes have 20%–25% sequence identity with microbial photolyases and40–60% sequence identity with (6-4) photolyase; and the sequence identity be-tween CRYs from various sources is remarkably high: human,Drosophila, andArabidopsiscryptochromes have about 60% sequence identity (54). However,CRYs from different sources have nonhomologous C-terminal extensions rangingfrom very short ones inDrosophilato as many as 240 amino acids in plants. Thetwo human CRYs are 73% homologous to each other but exhibit no sequencehomology within the C-terminal 75 amino acids. It is thought that this domainmay bind to effector molecules (14, 54). The hCRY1 protein is 586 amino acids inlength and encodes a protein of 66 kDa. The hCRY2 protein is 593 amino acids inlength and has a mass of 67 kDa (6). The mouse CRY proteins are nearly identicalto the human proteins with the exception of a short region in the C-terminal domain(7, 123).

At present there is no crystal structure of a cryptochrome. However, the crystalstructures of two photolyases (E. coliandA. nidulans) have been solved (124, 125).Even though these two enzymes share only 30% sequence identity and use dif-ferent second chromophores (a pterin versus a deazaflavin), the two structures arevirtually superimposable. Hence it is reasonable to assume that the human cryp-tochrome structure would be very similar to that ofE. coli photolyase (Figure 5).The enzyme has an overall dimension of 80× 60× 30 A and consists of twowell-defined domains interconnected by a loop of 62 amino acids. The N-terminalα/β domain adopts a fold (the Rossman fold) typical for dinucleotide binding do-mains although the primary sequence has no homology to other flavoproteins withthe Rossman fold and this domain does not actually bind the FAD dinucleotidecofactor of the enzyme. The pterin cofactor is located in the interdomain cleft nearthe surface and makes intimate contact with residues in theα/β domain. The C-terminal helical domain is made almost entirely ofα helices and resembles a slab of60× 40× 20 A.

The most prominent feature of theα/β domain is a hole in the center of theflat face that leads to FAD in the bottom of the hole. The dimensions of the holematch those of a pyrimidine dimer. It has been proposed that the enzyme flipsout the pyrimidine dimer from the DNA helix into the hole in close proximitywith the catalytic flavin cofactor so that photoinduced electron transfer and re-pair can occur with high efficiency (124). The FAD in photolyase has a U-shaped(or cis) conformation with the isoalloxazine ring and adenine in close proximity.This conformation is not found in flavoproteins, which carry out redox reactionsfrom the ground state, and it is thought to be unique to proteins of the photolyase/blue-light photoreceptor family, which carry out catalysis from an excited state(124).

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 49

Figure 5 Structure of photolyase. The crystal structure ofE. coli photolyase is shownin two forms. (Left) Ribbon diagram. Note the location of the MTHF photoantenna in thecrevice between theα/β domain and theα-helical domain and exposed to the solvent. TheFADH– catalytic cofactor is incisconformation (the adenine ring stacked on the flavin ring)and is deeply buried within theα-helical domain. The center-to-center distance between thetwo chromophores is 17A. (Right) Electrostatic surface potential and solvent accessiblesurface of the enzyme. Note the positively charged groove running the length of the moleculeand the hole (boxed) in the center of the groove; the hole leads to the FADH– in the core ofthe protein.

Remarkably, molecular modeling of hCRY1 and hCRY2 onto theα backbone ofE. coliphotolyase indicate that the overall geometry ofE. coliphotolyase, includingthe hole in the center, is retained in these proteins. These observations raise theinteresting possibility that cryptochromes have retained the unique features ofphotolyase reaction mechanism: dinucleotide flipping and photoinduced electrontransfer.

The human cryptochromes contain both chromophore/cofactors of the pho-tolyase/cryptochrome family of proteins. When expressed as recombinant pro-teins, they have identical near UV–blue absorption spectra with a 420-nm peak(6). However, cryptochromes purified from their natural sources are not currentlyavailable, and hence it is not known if the absorption spectra of the recombinantproteins are identical to those of the native proteins. The cofactors of this class ofproteins are readily lost or oxidized during purification from the natural sources(37, 66). Thus, the active form of photolyase contains the flavin in the two-electronreduced FADH– (or FADH2) form, but during purification the flavin in all but theS. cerevisiaephotolyase (58) becomes oxidized to the catalytically inert flavinneutral radical (FADHo) and fully oxidized FAD (57, 66). Thus, caution must beexercised in comparing the absorption spectra of purified cryptochromes to thecircadian action spectra.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

50 SANCAR

Expression of Cryptochromes in the Retinohypothalamic Axis

In humans, cryptochromes are expressed in all organs and hence the expressionpattern is not particularly revealing of function. It is noteworthy that the expressionof other genes thought to be strictly clock genes is not confined to the two organsknown to be critical for the circadian mechanism, the retina and the SCN. Thus,CLOCK, BMAL1, PER1, PER2, PER3, andTIM are expressed throughout thebody (126–132). However, histological examination of the expression ofmCry1andmCry2in the mouse retina and the brain is quite informative (7). The visualphotoreceptors, rhodopsin and the color opsins, are expressed in the inner segmentand to a lesser degree the outer segment of the retina. In contrast,mCry1andmCry2are expressed almost exclusively in the ganglion cell layer (GCL) and inner nuclearlayer (INL) and are evenly distributed in the central and peripheral retina (Figure 6).These are the layers that remain intact inrd mice, which are visually blind but havenormal circadian photoresponse (93). Both genes are also expressed throughoutthe brain. In general,mCry2 transcript is more abundant than that ofmCry1 inthe brain. In the pyramidal cell layer of the hippocampus, the granular cell layer

Figure 6 Expression ofmCry1 and mCry2 in mouse retina. In situ hybridization wasperformed with probes for cryptochromes and for the mouse opsin gene. BothCry genesare expressed in the ganglion cell layer (GCL) in a subset of ganglion cells indicated byarrows and in the inner nuclear layer (INL), whereas the opsin is expressed in the innersegment (IS) and the outer nuclear layer (ONL). The last two frames show the retinas ofa control mouse (right) and an agedrd mouse (left) in which only the regions expressingcryptochromes remain intact. (From References 7, 93.)

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 51

Figure 7 Expression ofmCry1andmCry2 in the mouse brain. Coronal brain sectionsmade at thezeitgebertime of ZT= 6 (see Figure 8 for azeitgeberscale) were probed withappropriate antisense RNAs. Both genes are highly expressed in all cerebral cortical layersbut particularly in the pyramidal cell layer of the hippocampus (H), the granular cell layer ofthe dentate gyrus (DG), and the pyramidal cell layer of the piriform cortex (PFC). Strikingly,the strongest signal ofmCry1 is in the SCN, wheremCry2expression is modest. (FromReference 7.)

of the dentate gyrus, and the pyramidal cell layer of the piriform cortex, the ratioof mCry2to mCry1transcript is about 2 to 1. However,mCry1 is expressed at ahigh level in the SCN, whereasmCry2expression in this region is almost negligi-ble (Figure 7). These expression patterns suggest that CRY1 and CRY2 performpartly redundant and partly complementary functions in running the circadianclock.

Circadian Oscillation of Cryptochrome Expression

The transcription of themCry1gene in the SCN exhibits a circadian oscillation(7) comparable to those of other circadian genes includingmPer1 (127, 128),mPer2 (133, 134),mPer3 (129, 135), andmTim (130, 136). ThemCry1 mRNAlevel reaches a maximum at ZT 6–8 (ZT= 0 by convention, the time at which thelight is turned on) and declines to a minimum at ZT 24 (Figure 8). The oscillationof mCry1 transcript persists when the mice are kept in constant darkness (137),as is observed for themPergenes and as is expected of a true circadian regulator.mPer1andmPer2, but notmPer3, can be induced from a low at nighttime to a highlevel of expression with light pulses of 5- to 60-min duration (133, 134).mCry1behaves likemPer3in that its expression is not inducible with acute light pulses(137). Because light pulses that inducemPer1andmPer2cause a proportional shiftin the phase of locomotor activity, it appears that acute phase shifting can occurwithout changing the circadian expression pattern ofmCry1in the SCN.

mCry1expression also shows circadian oscillation in internal organs, most no-tably in the liver (Figure 8); however, the phase of expression in internal organs isdelayed relative to the phase of SCN expression by about 8 h (7, 137, 144). Sim-ilar observations were made formPer1, mPer2, andmPer3(129). Taken together

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

52 SANCAR

Figure 8 Circadian oscillation ofmCry1expression in the SCN and liver. ThemCry1RNA levels were quantified by in situ hybridization in the SCN and RNase protection inthe liver. The values are expressed relative to the maximum within each set. (Adapted fromReferences 7, 137.)

these data constitute a strong case for the presence of peripheral clocks that arepresumably subordinate to the master clock in the SCN. Interestingly, even thoughmCry1is expressed at a high level in the testis, there is no obvious transcriptionaloscillation ofmCry1mRNA in this organ (137). Furthermore, themCry1expres-sion in testis seems to be restricted to spermatogonia with little or no expressionin Leydig cells (137). It has been reported that during meiosis, which occurs inthe development of spermatocytes, many genes are turned on but the transcriptsfrom these genes are not translated (138). This is not the case formCry1. In fact,testicular tissue contains the highest relative amount of mCRY1 protein of anytissue tested. However, as mice lackingCry1have apparently normal reproductivefunctions (139, 140), the significance of such high levels of mCRY1 in testis hasyet to be determined.

Cellular Localization of Cryptochromes

All known animal clock proteins are located either in the nucleus or shuttle betweenthe cytoplasm and nucleus to exert their feedback into the circadian transcrip-tion loop. Thus, the location of a potential photoreceptor molecule is importantfrom the standpoint of the phototransduction mechanism. A membrane-bound

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 53

photoreceptor such as rhodopsin must transduce the light signal through inter-mediate molecules to the effector clock proteins or genes located in the nucleus(12, 13). In contrast, a nuclear photoreceptor can interact directly with the clockgenes and proteins and thus function both as a photoreceptor and phototransducer.Cryptochromes appear to function by the latter mechanism because the followinglines of evidence show that they are located in the nucleus. First, yeast two-hybridanalysis revealed that CRY1 and CRY2 interact with the nuclear serine/threoninephosphatase 5 (141, 142). Second, transient transfection of human cells with full-lengthhCRY1-GFPandhCRY2-GFPfusion genes lead to nuclear accumulationof the fusion proteins as revealed by fluorescence microscopy (79, 143). A reportof mitochondrial localization of mCRY1 protein (123) has not been confirmed(143, 144). Third, both mCRY1 and mCRY2 bind to and affect the activities ofthe PER1 and PER2 proteins, which are known to shuttle from the cytoplasmto the nucleus (144). Finally, mCRY2 has the bipartite nuclear localization sig-nal PKRK-X13-KRAR in its C-terminal domain (79, 123). The fact that CRY1and CRY2 are essential components of the molecular clock, which resides in thenucleus, is further evidence for their nuclear localization.

Interactions of Cryptochromes with Other Clock Proteins

A yeast two-hybrid screen with the C-terminal domain of 381 amino acids, whichincludes the flavin binding site of hCRY2, detected the protein serine/threoninephosphatase 5 (PP5) as a hCRY2-interacting protein (141). PP5 is the only knownnuclear serine/threonine phosphatase in mammalian cells (142). It contains threetetratricopeptide repeats (TPR) and it binds to both hCRY1 and hCRY2 through thismotif, which is found in many proteins as a protein-protein interactionmotif. CRY proteins do not have the TPR motif. Binding of hCRY2 to PP5 reducesthe phosphatase activity by 80% in a light-independent manner (141). However,these experiments were conducted with recombinant hCRY2, which is severelydepleted of FAD, and this may explain why the effect of hCRY2 on PP5 was notaffected by light. Hence the physiological significance of CRY-PP5 interactionremains to be determined. However, it is known that inDrosophilaphosphoryla-tion/dephosphorylation of TIM and PER is an integral part of the photoentrainmentmechanism (12, 145, 146).

An in vitro assay with immobilized hCRY2 revealed specific interaction withPER1 and TIM1 but not CLOCK (139). A detailed study using a cotransfec-tion/coimmunoprecipitation assay coupled with reporter gene assay in transienttransfection experiments in NIH-3T3 cells showed the following (144): (a) BothmCRY1 and mCRY2 bind to mPER1, mPER2, mPER3, and mTIM; (b) CRY-PER interaction results either in nuclear transport or nuclear retention of PERs;(c) mCRY1 and mCRY2 inhibit transcription of a clock gene (mPer1) and anoutput gene (vasopressin) mediated by CLOCK•BMAL1 and MOP4•BMAL1,respectively. However, another study using both the yeast two-hybrid assay, and aCLOCK•BMAL1-driven luciferase assay with themPer1promoter as the target,

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

54 SANCAR

identified PER2, BMAL1, and CLOCK as the main targets of CRY1 and CRY2(146a). It was found that both CRY1 and CRY2 bound to PER2 strongly andthat CRY1 bound to BMAL1 whereas CRY2 bound to CLOCK with high affin-ity. Both CRY1 and CRY2 inhibited the transactivation ofmPer1promoter by theCLOCK•BMAL1 complex. There were some quantitative differences between theeffects of hCRY1 and hCRY2 on CLOCK•BMAL1 activity, which may explainthe differential effects ofmCRY1andmCRY2knockouts on circadian and molecu-lar behavior of the mutant mice addressed below. Significantly, in this study neitherthe interactions of CRY1 and CRY2 with the clock components, nor their negativeregulatory effects were affectd by light (146a). Clearly, the light-dependent func-tion of mammalian cryptochromes needs to be addressed by more direct assayswith native proteins and under physiological conditions.

Of potential relevance to the CRY photoreception/phototransduction mech-anism is the finding that inArabidopsisall morphogenetic phototransductionreactions depend on COP9 signalosome, which is related to the 26S proteasomecomplex (147, 148). Interestingly, the subunits of the COP9 signalosome are con-served between plants and animals, and the complex is located in the nucleuswhere it can directly interact with the molecular clockwork. However, at presentno evidence links the COP9 signalosome to the mammalian clockwork.

GENETICS OF MAMMALIAN CRYPTOCHROMES

No human or mouse genetic diseases are known to be caused by mutations incryptochrome genes. However, the biochemical and photochemical properties ofmammalian cryptochromes, the photoreceptor function of all other members ofthis family of proteins [photolyase, (6-4) photolyase, plant cryptochromes], and thecircadian oscillation ofmCry1in the SCN of mice constitute compelling evidencefor the involvement of cryptochromes in circadian photoreception and regulation.Direct evidence for the role ofCry genes in circadian regulation has been obtainedfrom the characteristics ofCry knockout mice lackingCry1, Cry2, or both.

Phenotype of Cry Mutant Mice

Cry1 knockout mice have seemingly normal circadian behavior under a regimenof 12 h of light and 12 h of darkness (LD12:12) (Figure 9), as tested by locomotor(wheel-running) activity (139, 140). In constant darkness, however, the animalsexhibited a free-running clock with 22.7-h periodicity, which is about 1 h shorterthan normal (Figure 9B, E). These data are consistent with an integral role ofmCRY1 protein in the transcription loop that generates the molecular clock.

Cry2 knockout mice show near-normal locomotor activity under the LD12:12condition but exhibit a free-running period 1 h longer than wild-type animals (79)when kept in darkness (Figure 9C, E). Thus, both mCRY1 and mCRY2 are impor-tant for maintaining a circadian clock of normal periodicity independent of their

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 55

Figure 9 Effect of mutations inmCry genes on locomotor (wheel-running) activityrhythms. (A–D) Wheel-running activity records of individual mice in the conventionaldouble-plotted format.Black represents times of activity. The animals were kept under anLD12:12 cycle and were transferred to constant darkness (DD) on the day indicated by anarrow. (A) Wild-type mouse. (B) Cry1–/– mouse. (C) Cry2–/– mouse. (D) Cry1–/–Cry2–/–

mouse. (E) Effect on circadian period of disrupting theCry genes. The free-running pe-riod is about 0.8 h shorter and 1.0 h longer than normal inCry1–/– andCry2–/– animals,respectively. The double knockout is arrythmic. (F) Fourier analyses of activity recordsshow no periodicity, circadian or otherwise, in the double knockout and unstable circadianperiodicity in theCry2–/– mutant. (From Reference 139.)

photoreceptor function (79, 139, 140). Interestingly, the response ofCry2–/– miceto acute light pulses was somewhat paradoxical. In wild-type mice, acute lightpulses given in the early hours of the morning cause an approximately 2-h phasedelay on the following day as measured by the onset of locomotor activity. Atface value one would expect that if the level of the circadian photoreceptor in theretina is reduced because of theCry2 knockout, the mutant mice would be lesssensitive to phase shifting. In fact, a light pulse delivered at CT17 (CT12 beingdefined as the onset of activity in mice under a DD [constant darkness] regimen)caused a 7-h phase delay inCry2–/– mice compared to 2 h inwild-type mice (79).This suggests that theCry2 gene dampens the effect of the photic input on theamplitude of the circadian oscillator (79). Indeed, as measured by the amplitudeof mPer1andmPer2mRNA oscillation and the length of the free-running periodof locomotor activity,mCRY1 andmCRY2 seem to pull the circadian oscilla-tor in opposite directions (139). However, the precise molecular mechanism ofthis unexpected effect ofCry2 on the mouse circadian behavior remains to beelucidated.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

56 SANCAR

The Cry1–/–Cry2–/– mice appear to have normal circadian locomotor activityunder an LD12:12 regimen; however, they instantly become arrhythmic uponswitching to DD (Figure 9D, E, F), indicating a complete collapse of the molecularoscillator under these conditions (139, 140). In fact, even under the LD12:12 regi-men, the seemingly entrained behavior might be caused by the “masking” activityof light (140), which is due to a decrease in activity in a nocturnal animal in adirect (visual) response to light (149, 150). These behavioral studies, then, showthat mCRY1 and mCRY2 proteins are core components of the circadian clock butdo not affect the visual (imaging) pathway of light perception. Interestingly, thecomplete collapse of the clock mechanism in the double mutant makes it impos-sible to assess by behavioral tests whether cryptochromes act as photoreceptors insynchronizing the circadian clock. This question was addressed at the molecularlevel by examining the effect of light onmPer1andmPer2transcription in mutantmice.

Status of the Molecular Clock in Cryptochrome Mutant Mice

Currently, the known components of the molecular oscillator in mice are CLOCK(126, 151), BMAL1 (131, 132), PER1, PER2, PER3 (127–129, 132–135), and TIM(129, 130, 136). The expression patterns ofPer1, 2, and3 show robust circadianoscillation in the SCN and peripheral tissues. In animals under an LD12:12 cycle,the expression of all threePer genes in the SCN increases during the light period(ZT 4–8) and declines at night, reaching a minimum at ZT 18–20. If the animalsare exposed to acute light pulses during the time of minimal expression (night orsubjective night), the transcriptions ofmPer1andmPer2are induced to daytimelevels (133, 134). Hence, the oscillation ofmPer1andmPer2transcription andtheir inducibility with acute light pulses were analyzed in cryptochrome mutantmice to evaluate the role of cryptochromes in photoreception (79, 139).

In Cry1–/– andCry2–/– mice, the inducibility ofmPer1is severely blunted. InCry1–/–Cry2–/– mice, there is no induction by acute light pulses (139), althoughin the double knockoutmPer1 is expressed at a high basal level at all times(Figure 10A). These data are consistent with mCRY1 and mCRY2 being the solephotoreceptors formPer1photoregulation as well as light-independent regulatorsof mPer1transcription. This interpretation is based on the fact that in the absenceof cryptochromesmPer1is expressed at near-maximal levels at all times, includingZT 18 whenmPer1transcription is normally at a minimum (139).

Surprisingly, the transcriptional behavior ofmPer2in the SCN of theCry mu-tant mice (139) is radically different from that ofmPer1(Figure 10B). First, underLD12:12 conditionsmPer2transcription oscillates with higher than wild-type am-plitude in bothCry1–/– andCry1–/–Cry2–/– mutants. Second, an acute light pulse atZT 18 inducesmPer2transcription to higher than wild-type levels in bothCry1–/–

andCry1–/–Cry2–/–mice. Finally, under all conditions the levels ofmPer2transcriptin the SCN of the mutant is higher than that of the wild type. These results areconsistent withmPer2transcription being negatively regulated by cryptochromes

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 57

Figure 10 Effects of disruption ofCrygenes on diurnal and circadian expression ofmPer1andmPer2in the SCN. The expression level was determined by in situ hybridization. (A, B)Diurnal expression patterns ofmPer1andmPer2and acute light induction ofmPergenesby a light pulse in the dark phase (hatched). ThemPer2oscillation was not tested inCry2–/–

animals. (C, D) Circadian expression patterns ofmPergenes under free-running conditions.BothCry1–/–andCry2–/–(not shown) mice exhibit circadian oscillation ofmPer1andmPer2expression although the phases of peak activities are different because of different periodlengths in the mutants. TheCry1–/–Cry2–/– animals express constitutively high levels ofmPer1and intermediate levels ofmPer2with no circadian oscillation of either, consistentwith the arrythmic behavior of these mice. (From References 79, 139.)

in a light-independent manner and, more significantly, indicate that there is acryptochrome-independent photic input pathway for induction ofmPer2transcrip-tion (139). At this point it is unclear whether themPer2oscillation under LD andits induction by light pulses are mediated by another circadian photoreceptor orby the visual system through a masking mechanism. Interestingly, under constantdarkness neithermPer1normPer2oscillate in the SCN of theCry1–/–Cry2–/– mice(Figure 10C, D), consistent with the arrhythmic behavior of these animals undera DD condition (Figure 9D).

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

58 SANCAR

Cryptochrome Genetics in Other Animals

Cryptochrome genes have been found in zebrafish,Xenopus laevis, andDrosophila(152–157) but, significantly, not inC. elegans. The Drosophila cryptochromehas been characterized in some detail. There is only one cryptochrome gene inDrosophilaand it is clearly involved in circadian photoreception (153, 156, 157).However, it is unclear at present whether photic input into the circadian clockcomes from the visual photoreceptor system (i.e. rhodopsin) in addition to cryp-tochrome. ADrosophilamutant was identified by its failure to expressPer in acyclic manner under LD conditions and it was found to have a mutation in thedCry gene (153). Furthermore, based on the crystal structure of DNA photolyase(124), this CRY mutant (CRYb) has an amino acid alteration in one of the residuesinvolved in binding to FAD (D410N). In theDrosophila Crymutants, PER andTIM are expressed at constitutive levels in the eye and the rest of the body withno detectable oscillation under either LD or DD conditions; in contrast, wild-typeanimals show robust circadian oscillation of both PER and TIM proteins. Sur-prisingly, however, the mutants were photoentrainable. To explain this paradoxthe expression of PER and TIM was examined in the lateral neuron (LN) cluster,which is the circadian pacemaker inDrosophila. PER and TIM oscillated weaklyin these neurons, which might explain the normal circadian behavior of the mutantflies.

The nearly normal photoentrainment of theCryb mutant was ascribed to photicinput from the visual (rhodopsin) photoreceptor. Surprisingly, a double mutant(norpA Cryb) that is totally deficient in the visual phototransduction system andpresumably in cryptochrome function was still photoentrainable at high light in-tensities but became unresponsive with low-intensity light (153). As no third pho-toreceptor is known inDrosophila, there is no satisfactory explanation of the data.One possibility is that since the mutation in CRY is conservative (Asp to Asn)Cryb does not have a null phenotype and the residual photoresponse in the dou-ble mutant is mediated by the mutant CRY photoreceptor. This may also explainwhy cryptochromeless mice are arrhythmic but theCryb homozygotes retain nor-mal circadian rhythm under DD: Even though CRYb is severely defective as aphotoreceptor, it may be capable of carrying out nearly normal clock function,which requires only protein-protein interactions. However, it is also possible thatthere are fundamental differences between the insect and mammalian circadianphotoregulatory systems.

TheDrosophilaCRY, like its mammalian counterparts, binds to TIM. Impor-tantly, this binding appears to be enhanced by light when tested by the yeasttwo-hybrid system, and light stimulated binding was reported to sequester TIMand thus induce CLOCK•BMAL1-controlled transcription of thePer gene in acotransfection experiment (158). This effect is opposite to the light-independentinhibition of gene transcription in mammalian cells (144, 146a) but parallels thelight-dependent andmCry1- andmCry2-dependent induction ofmPer1inductionin mice (79, 139).

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 59

MOLECULAR MODEL FOR THE MAMMALIANCIRCADIAN CLOCK

Evidence from mice to cyanobacteria indicate that circadian rhythms at the organ-ism level are engendered by a cell-autonomous and autoregulatory transcriptionloop (12, 13, 15, 111, 159). The main feature that distinguishes the circadian neg-ative feedback loop from other biochemical feedback loops is that the circadiansystem has a time delay between the transcription of the “clock gene” and theproduction or availability of the negative feedback inhibitory protein. This is ac-complished either by a delay of the translation of the transcript, a delay in activationby posttranslational modification, or a delay in the entry of the feedback inhibitorinto the nucleus (12, 13, 15). Many components of the circadian clock have beenidentified and characterized, so a reasonable model for the circadian clock and thecircadian system can now be constructed (Figure 11).

The cryptochromes are the photoreceptors and essential components of thecircadian oscillator, performing partly redundant and partly complementary func-tions. Currently it is not known how the cryptochromes transmit the light signalto the molecular clock and how they engender a nerve impulse transmitted to thebrain. In contrast, there is a reasonably detailed understanding of the molecularclock: CLOCK and BMAL1 are positive transcription factors, which dimerize,bind to the E-box of and turn on thePer andTim genes (131, 132). There is adelay between the transcription and translation of these genes such that significantaccumulation of PER1 and TIM in the nucleus is delayed by about 6 h from thetime of maximum transcription (160). When PER1, PER2, and PER3 enter the nu-cleus, they compete with BMAL1 for heterodimerization with CLOCK using thePAS dimerization domain. The CLOCK•PER, CLOCK•TIM or BMAL1 •PER,BMAL1•TIM heterodimers are inactive as transcription factors. Alternatively, ashas been shown inDrosophila, PER and TIM may bind to the CLOCK•BMAL1heterodimer and interfere with its activity (161). As a consequence, the transcrip-tion of Per andTim is turned off and the corresponding proteins are degraded,allowing the formation of the CLOCK•BMAL1 complex and initiation of a newcycle. CRY1 and CRY2 proteins play a central role in this cycle, independent oftheir photoreceptor function, by directly interacting with CLOCK, BMAL1, PER,and TIM proteins, transporting or retaining the PER proteins in the nucleus. In do-ing so they interfere with the CLOCK•BMAL1 transcriptional activator and thusinhibit the transcription ofPer genes and other clock genes. This model is by ne-cessity general and qualitative because all the components of the molecular clockhave not yet been identified, and the precise roles and mechanisms of action of theknown components have not yet been elucidated. In particular, the mechanismsof photoreception and signal phototransduction (most likely by electron transfer)by cryptochromes must be clarified in order to construct a more specific model.Nevertheless, the working model is likely to be correct in outline and should serveas a guide for testing and developing more accurate models.

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

60 SANCAR

Figure 11 Molecular model for the mammalian circadian clock. The clock is locatedwithin the nucleus. The positive and negative regulatory effects of the circadian loop areindicated bygreenandred arrows, respectively. The CLOCK•BMAL1 heterodimer is atranscriptional activator that turns on themPergenes. The PER proteins enter the nucleuseither as PER heterodimers or as PER•CRY heterodimers and bind to CLOCK•BMAL1to interfere with its activator functions either as such or as PER•TIM heterodimers andthus complete the circadian autoregulatory loop. Cryptochromes mediate light inductionof mPer1but notmPer2, but they may positively affect the steady-state cycling of bothPer genes (left). In addition, the cryptochromes dimerize with PERs, and perhaps directlyinteract with CLOCK•BMAL1 and thus function in the negative feedback loop as well(right). Finally, differences in the relative contributions ofCRY1andCRY2in the positivedrive and negative feedback components of the loop must account for the differences intheir influence on circadian period. (Adapted from Reference 139.)

CRYPTOCHROMES AND HUMAN HEALTH

The following diseases and syndromes are known or thought to be caused by abnor-mal clock function. To what degree, if any, abnormal functioning of cryptochromescontribute to pathogenesis of these conditions is not known at present.

Seasonal Affective Disorder

Seasonal affective disorder (SAD) may affect up to 5% of the population, whomanifest classic symptoms of depression including withdrawal, morbid thoughts,sadness, decreased activity, and loss of libido and some atypical features such as

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 61

overeating, weight gain, and hypersomnia (18). It occurs during the winter monthsin the Northern Hemisphere. Its pathogenesis is not understood. However, up to75% of the patients go into remission with daily exposure to high-intensity whitelight for 30 min (162). This is consistent with an abnormality in the circadian systemand conceivably with mutations in the clock components including cryptochromes,or even with vitamin B2 and folic acid deficiency, which results in a lower thannormal level of active cryptochromes.

Delayed Sleep Phase Syndrome

Patients with delayed sleep phase syndrome typically go to sleep around 4:00AM and wake up around noon. A substantial number of these patients benefitfrom phase shifting with intense light exposure of short duration. Interestingly,some patients benefit from vitamin B12 given at high doses (163). It is conceivablethat some of these patients carry a missense mutation in one of the cryptochromegenes.

Jet Lag (Syndrome of Rapid Change in Time Zone)

Changing time zones in a short period imposes a new phase-shifted dark-lightcycle on the traveler. Until the individual’s internal clock synchronizes to the newenvironment, the social demands of the new time zone and the person’s capabilityfor fulfilling them are out of phase. As a consequence, the person suffers frominsomnia, fatigue, and irritability. As a rule it takes about one day for each timezone change to synchronize the biological rhythm with the new environment (18).In mammals including humans, the pineal hormone melatonin reaches its maxi-mum during the night, and its secretion can be suppressed by exposure to acutelight pulses. The SCN contains high-affinity melatonin receptors (164, 165), andtreatment with melatonin has been used to remedy the effects of jet lag. However,even though the drug can entrain locomotor activity in hamsters (164), its useful-ness in humans for treating jet lag and a host of other ailments is controversial(165).

Rotating Shift Work

Shift work can cause sleep disturbances and fatigue, and impair mental and physicalperformance. The syndrome is caused by the discrepancy between social demandon the individual to perform certain tasks and the individual’s ability to performthese tasks optimally as dictated by the circadian clock. A person working thenight shift cannot perform at peak mental and physical levels. It has been suggestedthat the disasters at Three Mile Island, Chernobyl, and Bhopal occurred betweenmidnight and dawn at least in part because of errors committed by workers whohad not been synchronized to work the night shift (166). The symptoms can beameliorated by exposure to high-intensity light during the night to shift the phaseof the circadian clock (167).

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

62 SANCAR

Circadian Clock and Breast Cancer

In industrialized countries the incidence of breast cancer has steadily increased dur-ing this century. It has been suggested that one cause is exposure to light for longerperiods than that afforded by the natural daily light-dark cycle. This so-calledmelatonin hypothesis proposes that the suppression of melatonin secretion at nightby artificial light increases breast cancer risk by increasing exposure to estrogen(168, 169). Since the isolation of melatonin receptors and the finding of widespreadexpression of these receptors in neural and nonneural tissues (170, 171), the scopeof the melatonin hypothesis has expanded to include other mechanisms by whichreduced melatonin concentrations may affect hormone homeostasis and inducebreast and perhaps other cancers (172). A retrospective survey found that severelyblind women had about half the incidence of breast cancer of women with normalvision (173). A recent study that compared the breast cancer incidence in womenwith various degrees of visual impairment to that in women with normal visionfound that women with no conscious perception of light had a 60% lower incidenceof breast cancer than women with normal vision. Women with less severe visualimpairment had intermediate incidence of breast cancer (174). Further studies areneeded to understand the contributions of the circadian and visual phototrans-duction systems to these phenomena and to provide a mechanistic basis for theepidemiological data.

CONCLUDING REMARKS

Since the discovery of the first mammalian circadian mutant in 1988 (175) and theisolation of the first mouse circadian mutant in 1994 (151), there has been con-siderable progress in the genetics and molecular biology of the circadian clock.At present eight human and mouse genes that qualify for the definition of “clockgene” have been isolated and characterized:Clock, BMal1, Per1, Per2, Per3, Tim,Cry1, andCry2. From the initial characterization of these genes and their proteinsa reasonably detailed molecular model for the mammalian circadian clock hasbeen developed that appears to be based, as in other species, on an autoregulatorytranscriptional loop.

Recent research has forced a conceptual change in the models for circadianrhythms. Whereas the classical circadian rhythm models consisted of three com-ponents—input, clock, and output—it is now clear that at the molecular levelthe input molecules (cryptochromes) are also basic components of the molecularclockwork. The same may also be true for some of the output molecules. It isexpected that in the near future the entire set of the mammalian clock genes willbe identified. This should make it possible to develop a cell-free system that oscil-lates with circadian periodicity and that can be reset by light through cryptochromeblue-light photoreceptors. The study of clock-setting by cryptochromes will un-doubtedly reveal a novel signal transduction mechanism whose significance maywell transcend the field of circadian research. Finally, understanding of circadian

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 63

photoreception and of the circadian clock at the molecular level should help inrational drug design for clock-related illnesses, and in development of rationalbehavioral approaches to enhance and optimize health, well-being, and physicaland intellectual performance.

ACKNOWLEDGMENTS

I wish to acknowledge the reviews by Presti & Delbr¨uck (22), Schwartz (18), andRonneberg & Foster (9), which guided me in organizing this review. I thank MHVitaterna, JS Takahashi, and T Todo for collaborative work on cryptochromes. Ithank C Thompson, M Snider, and C Brautigam for their help with the figures,C Thompson and C Brautigam for comments on the manuscript, and M Sanderfor professional scientific editing. This work was supported by the NIH grantGM31082.

Visit the Annual Reviews home page at www.AnnualReviews.org

LITERATURE CITED

1. Boll F. 1877.Arch. Anat. Physiol. Abt.42. Wald G. 1934.J. Gen. Physiol.18:905–

153. Wald G. 1935.J. Gen. Physiol.19:351–

714. Wald G. 1968.Science162:230–395. Nathans J, Thomas D, Hogness DS. 1986.

Science232:193–2026. Hsu DS, Zhao XD, Zhao SY, Kazantsev

A, Wang RP, et al. 1996.Biochemistry35:13871–77

7. Miyamoto Y, Sancar A. 1998.Proc. Natl.Acad. Sci. USA95:6097–102

8. Blakeslee S. 1998.New York TimesCXLVIII:F2 (Dec. 15)

9. Roenneberg T, Foster RG. 1997.Pho-tochem. Photobiol.66:549–61

10. Foster RG. 1998.Neuron20:829–3211. Lucas RJ, Foster RG. 1999.Curr. Biol.

9:214–1712. Young MW. 1998.Annu. Rev. Biochem.

67:135–5213. Dunlap JC. 1999.Cell 96:271–9014. Cashmore AR, Jarillo JA, Wu YJ, Liu D.

1999.Science284:760–6515. Takahashi JS. 1995.Annu. Rev. Neurosci.

18:531–33

16. Ouyang Y, Andersson CR, Kondo T,Golden SS, Johnson CA. 1998.Proc. Natl.Acad. Sci. USA95:8660–64

17. DeCoursey PJ, Krulas JR, Mele G,Holley DC. 1997. Physiol. Behav.62:1099–108

18. Schwartz WJ. 1993.Adv. Intern. Med.38:81–106

19. Czeisler CA, Duffy JF, Shanahan TL,Brown EN, Mitchell JF, et al. 1999.Sci-ence284:2177–81

20. Winfree A. 1972.J. Insect Physiol.18:181–85

21. Liu Y, Merrow M, Loros JJ, Dunlap JC.1998.Science281:825–29

22. Presti D, Delbr¨uck M. 1978.Plant Cell En-viron. 1:81–100

23. Song PS, Moore TA. 1974.Photochem.Photobiol.19:435–41

24. Sancar A, Sancar GB. 1984.J. Mol. Biol.172:223–27

25. Eker APM, Hessels JKC, van deVelde J. 1988.Biochemistry 27:1758–65

26. Todo T, Kim ST, Hitomi K, Otoshi E, InuiT, et al. 1997.Nucleic Acids Res.25:764–68

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

64 SANCAR

27. Hitomi K, Kim ST, Iwai S, Narima N,Otoshi E, et al. 1997.J. Biol. Chem.272:32591–98

28. Zhao XD, Liu JQ, Hsu DS, Zhao SY, Tay-lor JS, Sancar A. 1997.J. Biol. Chem.272:32580–90

29. Malhotra K, Kim ST, Batschauer A, DawutL, Sancar A. 1995.Biochemistry34:8892–99

30. Lin C, Robertson DE, Ahmad M, RaibekasAA, Jorns MS, et al. 1995.Science269:968–70

31. Christie JM, Reymond P, Powell GK,Bernasconi P, Raibekas AA, et al. 1998.Science282:1698–701

32. Christie JM, Salomon M, Nozue K, WadaM, Briggs WR. 1999.Proc. Natl. Acad. Sci.USA96:8779–83

33. Walsh C. 1986.Acc. Chem. Res.19:216–21

34. Eker APM, Kooiman P, Hessels JKC,Yasui A. 1990.J. Biol. Chem.265:8009–15

35. Malhotra K, Kim ST, Walsh CT, Sancar A.1992.J. Biol. Chem.267:15406–11

36. Kim ST, Heelis PF, Sancar A. 1992.Bio-chemistry31:11244–48

37. Johnson JL, Hamm-Alvarez S, PayneG, Sancar GB, Rajagopalan KV,Sancar A. 1988. Proc. Natl. Acad.Sci. USA85:2046–5037

38. Tsien RY. 1998.Annu. Rev. Biochem.67:509–44

39. Jiang ZY, Swem LR, Rushing BG,Devanathan S, Tollin G, Bauer CE. 1999.Science285:406–9

40. Deisenhofer J, Michel H. 1989.EMBO J.8:2149–70

41. Quail PH, Boylan MT, Parks BM, ShortTWE, Xu Y, Wagner D. 1995.Science268:675–80

42. Yeh KC, Lagarias JC. 1998.Proc. Natl.Acad. Sci. USA95:13976–81

43. Hughes J, Lamparter T, Mittmann F, Hart-mann E, Gartner W, et al. 1997.Nature386:663

44. Yeh KC, Wu SH, Murphy JT, Lagarias JC.1997.Science271:1505–8

45. Fankhauser C, Yeh KC, Lagarias JC,Zhang H, Elich TD, Chory J. 1999.Sci-ence284:1539–41

46. Hoecher U, Tepperman JM, Quail PH.1998.Science284:496–99

47. Ni M, Tepperman JM, Quail PH. 1998.Cell 95:657–69

47a. Ni M, Tepperman JM, Quail PH. 1999.Nature400:781–83

47b. Choi G, Yi H, Lee J, Kwan YK, Soh MS,et al. 1999.Nature401:610–13

48. Huala E, Oeller PW, Liscum E, HansIS, Larsen E, Briggs WR. 1997.Science278:2120–23

49. Nozue K, Kanegae T, Imaizumi T, FukudaS, Okamoto H, et al. 1998.Proc. Natl.Acad. Sci. USA95:15826–30

50. Sancar A. 1994.Biochemistry 33:2–7

51. Todo T, Takemori H, Ryo H, IharaM, Matsunaga T, et al. 1993.Nature361:371–74

52. Todo T, Ryo H, Yamamoto K, Toh H,Inui T, et al. 1996.Science272:109–12

53. Nakajima S, Sugiyama M, Iwai S,Hitomi K, Otoshi E, et al. 1998.NucleicAcids Res.26:638–44

54. Todo T. 1999.Mutat. Res.434:89–9754a. Afonso CL, Tulman ER, Lu Z, Ome

E, Kutish GF, Rock DL. 1999.J. Virol.73:533–52

55. Li YF, Kim ST, Sancar A. 1993.Proc.Natl. Acad. Sci. USA90:4389–93

56. Kiener A, Husain I, Sancar A, Walsh C.1989.J. Biol. Chem.264:13880–87

57. Sancar GB, Jorns MS, Payne G, Fluke DJ,Rupert CS, Sancar A. 1987.J. Biol. Chem.262:492–98

58. Sancar GB, Smith FW, Heelis PF. 1987.J. Biol. Chem.262:15457–65

59. Kim ST, Malhotra K, Smith CA, Tay-lor JS, Sancar A. 1994.J. Biol. Chem.269:8535–40

60. Gressel J. 1977.Photochem. Photobiol.30:749–54

61. Galland P, Senger H. 1988.Photochem.Photobiol.48:811–20

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 65

62. Short TW, Briggs WR. 1994.Annu. Rev.Plant Physiol. Plant Mol. Biol.45:143–71

63. Ninnemann H. 1995.Photochem. Photo-biol. 61:22–31

63a. Briggs WR, Huala E. 1999.Annu. Rev.Cell Dev. Biol.15:33–62

64. Sancar A, Rupert CS. 1978.Gene4:294–308

65. Sancar GB, Smith FW, Lorence MC,Rupert CS, Sancar A. 1984.J. Biol. Chem.259:6033–38

66. Jorns MS, Sancar GB, Sancar A. 1984.Biochemistry23:2673–79

67. Rupert CS, Goodgal SH, Herriott RM.1958.J. Gen. Physiol.41:451–71

68. Jagger J. 1958.Bacteriol. Rev.22:99–140

69. Ahmad M, Cashmore AR. 1993.Nature366:162–66

70. Adams MD, Kerlavage AR, FleischmannRD, Fuldner RA, Bult CJ, et al. 1995.Na-ture377:3–174

71. Batschauer A. 1993.Plant J.4:705–972. Hoffman PD, Batschauer A, Hays

JB. 1996.Mol. Gen. Genet.253:259–65

73. Lin CT, Ahmad M, Chan J, Cashmore AR.1996.Plant Physiol.110:1047

74. Guo HW, Yang WY, Mockler TC, Lin CT.1998.Science279:1360–63

75. Small GD, Min B, Lefebvre PA. 1995.Plant Mol. Biol.28:443–54

76. Kanegae T, Wada M. 1998.Mol. Gen.Genet.259:345–53

76a. Guo HW, Duong H, Ma N, Lin CT. 1999.Plant J.19:279–87

76b. Kleiner O, Kircher S, Harter K,Batschauer A. 1999.Plant J.19:289–96

77. Ahmad M, Jarillo JA, Smirnova O,Cashmore AR. 1998.Mol. Cell 1:939–48

78. Lin CT, Yang HY, Guo HW, Mockler T,Chen J, Cashmore AR. 1998.Proc. Natl.Acad. Sci. USA95:2686–90

79. Thresher RJ, Vitaterna MH, Miyamoto Y,Kazantsev A, Hsu DS, et al. 1998.Science282:1490–94

80. Somers DE, Devlin PF, Kay SA. 1998.Science282:1488–90

81. van der Spek PJ, Kobayashi K, BootsmaD, Takao M, Eker APM, Yasui A. 1996.Genomics87:177–82

82. Frank KD, Zimmerman WF. 1969.Sci-ence163:688–89

83. Klemm E, Ninnemann H. 1976.Pho-tochem. Photobiol.24:369–71

84. Suri VP, Qian ZW, Hall JC, Rosbash M.1998.Neuron21:225–34

85. Takahashi JS, DeCoursey P, BaumanL, Menaker M. 1984.Nature 308:186–88

86. Provencio I, Foster RG. 1995.Brain Res.694:183–90

87. Yoshimura T, Ebihara S. 1996.J. Comp.Physiol. A178:797–802

88. Dillon J, Zheng L, Merriam JC, Gail-lard ER. 2000.Photochem. Photobiol.71:225–29

89. Paietta J, Sargent ML. 1981.Proc. Natl.Acad. Sci. USA78:5573–77

90. Zimmerman WF, Goldsmith TH. 1971.Science171:1167–68

91. Yang ZH, Emerson M, Su HS, Sehgal A.1998.Neuron21:215–23

92. Bowes C, Li T, Danciger M, Baxter LC,Applebury ML, Farber DB. 1990.Nature347:677–80

93. Foster RG, Provencio I, Hudson D, FiskeS, DeGrip W, Menaker M. 1991.J. Comp.Physiol. A169:39–50

94. Yoshimura T, Ebihara S. 1998.Brain Res.779:188–93

95. Freedman MS, Lucas RJ, Soni B, vonSchantz M, Munoz M, et al. 1999.Sci-ence284:502–4

96. Lucas RJ, Freedman MS, Munoz M,Garcia-Fernandez JM, Foster RG. 1999.Science284:505–7

97. Lucas RJ, Foster RG. 1999.J. Biol.Rhythms14:4–9

98. Czeisler CA, Shanahan TL, Klerman EB,Martens H, Brotman DJ, et al. 1995.N.Engl. J. Med.332:6–11

99. Menaker M. 1968.Proc. Natl. Acad. Sci.USA59:414–21

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

66 SANCAR

100. Takahashi JS, Hamm H, Menaker M.1980. Proc. Natl. Acad. Sci. USA77:2319–22

101. Underwood H, Menaker M. 1970.Science170:190–93

102. Okano T, Yoshizawa T, Fukada Y. 1994.Nature372:94–97

103. Max M, McKinnon PJ, Seidenman KJ,Barrett RK, Applebury ML, et al. 1995.Science36:1502–6

104. Provencio I, Jiang GS, DeGrip WJ, HayesWP, Rollag MD. 1998.Proc. Natl. Acad.Sci. USA95:340–45

105. Soni BG, Foster RG. 1997.FEBS Lett.406:279–83

106. Soni BG, Philip AR, Foster RG, Knox BE.1998.Nature394:27–28

107. Blackshaw S, Snyder SH. 1999.J. Neu-rosci.19:3681–90

108. Green CB. 1998.Trends Cell Biol.8:224–30

109. Menaker M, Moreira LF, Tosini G. 1997.J. Med. Biol. Res.30:305–13

110. Hastings JW, Sweeney BM. 1960.J. Gen.Physiol.43:697–706

111. Johnson CH, Golden SS, Ishiura M,Kondo T. 1996.Mol. Microbiol. 21:5–11

112. Roenneberg T, Hastings JW. 1991.Pho-tochem. Photobiol.53:525–33

113. Campbell SS, Murphy PJ. 1998.Science279:396–99

114. Meijer JH, Thio B, Albus H, Schaap J,Ruijs ACJ. 1999.Brain Res.831:337–39

115. Schwartz WJ. 1996.Sci. Med.3:44–53116. Moore RY, Eichler VB. 1972.Brain Res.

42:201–6117. Giebultowicz JM, Hege DM. 1997.Na-

ture386:664118. Plautz JD, Kaneko M, Hall JC, Kay SA.

1997.Science278:1632–35119. Balsalobre A, Damiola F, Schibler U.

1998.Cell 93:929–37120. Tosini G, Menaker M. 1996.Science

272:419–21121. Earnest DE, Liang FQ, Ratcliff M,

Cassone VM. 1999.Science283:693–95

122. Sakamoto K, Nagase T, Fukui H,Horikawa K, Okada T, et al. 1998.J. Biol.Chem.273:27039–42

123. Kobayashi K, Kanno S, Smit B, van derHorst GTJ, Takao M, Yasui A. 1998.Nu-cleic Acids Res.26:5086–92

124. Park HW, Kim ST, Sancar A, DeisenhoferJ. 1995.Science268:1866–72

125. Tamada T, Kitadokoro K, Higuchi Y,Inaka K, Yasui A, et al. 1997.Nat. Struct.Biol. 4:887–91

126. King DP, Zhao YL, Sangoram AM, Wils-bacher LD, Tanaka M, et al. 1997.Cell89:641–53

127. Sun ZS, Albrecht U, Zhuchenko O,Bailey J, Eichele G, Lee CC. 1997.Cell90:1003–11

128. Tei H, Okamura H, Shigeyoshi Y,Fukuhara C, Ozawa R, et al. 1997.Na-ture389:512–16

129. Zylka MJ, Shearman LP, Weaver DR,Reppert SM. 1998.Neuron 20:1103–10

130. Sangoram AM, Saez L, Antoch MP,Gekakis N, Staknis D, et al. 1998.Neuron21:1101–13

131. Hogenesch JB, Gu YZ, Jain S, BradfieldCA. 1998. Proc. Natl. Acad. Sci. USA95:5474–79

132. Gekakis N, Staknis D, Nguyen HB, DavisFC, Wilsbacher LD, et al. 1998.Science280:1564–69

133. Albrecht U, Sun ZS, Eichele G, Lee CC.1997.Cell 91:1055–64

134. Shigeyoshi Y, Taguchi K, Yamamoto S,Takeida S, Yan L, et al. 1997.Cell91:1043–53

135. Takumi T, Taguchi K, Miyake S, SakakidaY, Tokashima N, et al. 1998.EMBO J.17:4753–59

136. Takumi T, Nagamine Y, Miyake S,Matsubara C, Taguchi K, et al. 1999.Genes Cells4:67–75

137. Miyamoto Y, Sancar A. 1999.Mol. Brain.Res.71:248–53

138. Eddy EM, O’Brien DA. 1998.Curr. Top.Dev. Biol.37:141–200

P1: FIB

July 11, 2000 17:11 AR102 CHAP02

CRYPTOCHROME CIRCADIAN PHOTORECEPTOR 67

139. Vitaterna MH, Selby CP, Todo T, NiwaH, Thompson C, et al. 1999.Proc. Natl.Acad. Sci. USA96:12114–19

140. van der Horst GTJ, Muijtjens M,Kobayashi K, Tokano R, Kanno S, et al.1999.Nature 348:627–30

141. Zhao SY, Sancar A. 1997.Photochem.Photobiol.66:727–31

142. Cohen PTW. 1997.Trends Biochem. Sci.22:245–51

143. Petit C, Sancar A. 1999.Med. Sci.15:1411–18

144. Kume K, Zylka MJ, Sriram S, Shear-man LP, Weaver DR, et al. 1999.Cell98:193–205

145. Zeng HK, Qian ZW, Myers MP, RosbashM. 1996.Nature380:129–35

146. Kloss B, Price JL, Saez L, Blau J, Ro-thenfluh A, et al. 1998.Cell 94:97–107

146a. Griffin EA, Staknis D, Weitz CJ. 1999.Science286:768–71

147. Wei N, Deng XW. 1998.Photochem.Photobiol.68:237–41

148. Seeger M, Kraft R, Ferrell K, Bech-Otschir D, Dumdey R, et al. 1998.FASEB J.12:469–78

149. Mikkelsen JD, Vrang N, Mrosovsky N.1998.Brain Res. Bull.47:367–76

150. Fleissner JF, Fleissner G. 1992.Disc.Neurosci.8:79–84

151. Vitaterna MH, King DP, Chang AM,Kornhauser JM, Lowrey PL, et al. 1994.Science264:719–25

152. Emery P, So WV, Kaneko M, Hall JC,Rosbash M. 1998.Cell 95:669–79

153. Stanevsky R, Kaneko M, Emery P,Peretta B, Wager-Smith K, et al. 1998.Cell 95:681–92

154. Selby CP, Sancar A. 1999.Photochem.Photobiol.69:105–7

155. Okano S, Kanno S, Takao M, Eker APM,Isono Y, et al. 1999.Photochem. Photo-biol. 69:108–13

156. Ishikawa T, Matsumoto A, Kato T,Togashi S, Ryo H, et al. 1999.GenesCells4:57–66

157. Egan EE, Franklin TM, Hildebrand-Chae MJ, McNeil GP, Roberts MA,et al. 1999.J. Neurosci.19:3665–73

158. Ceriani MF, Darlington TK, Staknis D,Mas P, Petti AA, et al. 1999.Science285:553–56

159. Herzog ED, Takahashi JS, Block GD.1998.Nat. Neurosci.1:708–13

160. Hastings MH, Field MD, Maywood ES,Weaver DR, Reppert SM. 1999.J. Neu-rosci.19:RC11(1–7)

161. Lee C, Bae K, Edery I. 1999.Mol. Cell.Biol. 19:5316–25

162. Wehr TA, Rosenthal NE. 1989.Am. J.Psychiatry146:829–39

163. Okawa M, Mishima K, Nanami T,Shimizu T, Iljima S, et al. 1990.Sleep13:15–23

164. Armstrong SM. 1989.Pineal Res. Rev.7:157–202

165. Reppert SM, Weaver DR. 1995.Cell83:1059–62

166. Folkard S. 1990.Philos. Trans. R. Soc.London Ser. B327:543–53

167. Dawson D, Campbell SS. 1991.Sleep14:511–16

168. Stevens RG, Davis S, Thomas DR,Anderson LE, Wilson BW. 1992.FASEBJ. 6:853–60

169. Stevens RG, Davis S. 1996.Environ.Health Perspect.104:135–40

170. Weaver DR, Rivkees SA, Reppert SM.1989.J. Neurosci.9:2581–90

171. Becker-Andre M, Wiesengerg I,Schaeren-Wiemers N, Andre E, Miss-bach M, et al. 1994.J. Biol. Chem.269:28531–34

172. Baldwin WS, Barrett JC. 1998.Mol.Carcinog.21:149–55

173. Hahn RA. 1991.Epidemiology2:208–13

174. Verkasalo PK, Pukkala E, Stevens RG,Ojamo M, Rudanko SL. 1999.Br. J.Cancer80:1459–60

175. Ralph MR, Menaker M. 1988.Science241:1225–27